Cocaine dependence is difficult to treat. Despite more than 40 controlled double-blind trials, no medication has yet been identified as effective for cocaine dependence. To date, most trials have employed heterogeneous samples in terms of baseline level and severity of cocaine use. The possibility exists that certain medications may be more or less effective depending on baseline cocaine use. We propose an innovative methodology that specifies baseline abstinence status a priori, and tests whether this variable moderates the relationship between pharmacotherapy and outcome. In this revised application we have selected three potential medications for which there exists a compelling rationale for examining differential treatment effects under baseline abstinent and nonabstinent conditions. Based on preliminary findings, we hypothesize that naltrexone 50 mg and levodopa/carbidopa 800/200 mg will show significant efficacy over placebo when administered on a baseline of cocaine abstinence, consistent with a "relapse prevention" mechanism. Further, we hypothesize that modafinil 400 mg will show significant efficacy over placebo when administered on a baseline of cocaine non-abstinence, consistent with an "abstinence initiation" mechanism. This 5-year project begins with a 4-week baseline phase involving motivational interviewing and incentive strategies for initiating abstinence in treatment-seeking cocaine users. The second phase of the study consists of a 12-week medication trial comparing four pharmacotherapy conditions (placebo, naltrexone 50mg, levodopa/carbidopa 800/200 mg, and modafinil 400 mg) with stratified random assignment to condition based on baseline abstinence status (abstinent/non-abstinent). During treatment all subjects (N=200) will receive psychotherapy and contingency management designed to enhance medication compliance, with urine screening conducted on a fixed, thrice-weekly schedule. Objective and self-reported drug use will continue to be measured after treatment (3-, 6-months). The assessment battery will capture data necessary to explore secondary aims of the projects, i.e., factors related to achievement of initial abstinence and possible treatment mechanisms by which different medications facilitate cocaine cessation or relapse prevention. Hierarchical multiple regression analyses will be used to determine whether baseline status moderates the effect of pharmacotherapy on outcome. For significant interactions, regression slopes for each treatment condition at the two levels of the moderator (abstinent vs non-abstinent) will be plotted. This project will contribute to our P50 medication development research program by identifying subpopulations and baseline conditions most responsive to treatment, thus providing unique new and valuable information to guide future treatment decision making.